Free Models
Neural Inverse Cloud hosts open-source AI models at free.neuralinverse.com. Free forever — no credit card, no API key, no trial expiry.
Requirements
Free Models are available when you're working inside a Neural Inverse Cloud workspace. This applies to:
- Cloud Web IDE — open your workspace at
cloud.neuralinverse.com, models work immediately - Desktop IDE connected to Cloud — connect via Remote Explorer → Neural Inverse Cloud
Free Models are not available in standalone local mode (no workspace connection). This is by design — the models are a benefit of the Neural Inverse Cloud platform.
Don't have an account? Sign up free at cloud.neuralinverse.com — no credit card required.
Available Models
| Model | Type |
|---|---|
deepseek-r1 | Reasoning |
deepseek-r1-0528 | Reasoning (updated checkpoint) |
deepseek-v3 | General purpose |
deepseek-v4-flash | Fast inference |
deepseek-v4-pro | Higher capability |
llama-3.3-70b | Meta Llama 3.3 |
llama-4-maverick | Meta Llama 4 |
mistral-large-3 | Mistral open-weight |
kimi-k2.6 | Moonshot AI open-weight |
These appear as niFreeModels in the model picker.
Getting Started
- Sign up at cloud.neuralinverse.com
- Create a workspace (any template)
- Open your workspace — via the web IDE or connect your desktop IDE through Remote Explorer
- Select a free model from the model picker in Power Mode or the Agent panel
No configuration needed. Models are available the moment your workspace is running.
Troubleshooting
"Neural Inverse Free Models requires a Neural Inverse Cloud workspace"
This error means you're running the IDE locally without a cloud workspace connection.
Fix: Connect to a Neural Inverse Cloud workspace via Remote Explorer → Neural Inverse Cloud. If you don't have an account, sign up free.
"Neural Inverse Free Models requires an active cloud session"
Your session token has expired.
Fix: Re-authenticate via the Neural Inverse Cloud extension in the sidebar.
Using Your Own Models Instead
If you prefer to use models without a cloud workspace, configure any provider under LLM Providers (BYOLLM). Ollama, vLLM, and LM Studio run entirely local — no account needed.
Third-Party Models
Claude, GPT, Gemini, and other proprietary models require your own API credentials. See Model Gateway for cloud-managed keys, or configure them directly in settings.
Self-Host
The free-models service is open source. Self-host it and point your IDE at your own instance via the niFreeModels endpoint setting.